Update app.py

app.py

@@ -85,7 +85,8 @@ ode_method = "euler"
 sway_sampling_coef = -1.0
 speed = 1
 fix_duration = None
-language = "en"
+ref_language = "en-us"
+language = "en-us"
 
 
 DEFAULT_TTS_MODEL = "F5-TTS"
@@ -262,7 +263,8 @@ def infer_batch(ref_audio, ref_text, gen_text_batches, exp_name, remove_silence,
 # Define weights for characters
     punctuation_weights = {",": 0, ".": 0, " ": 0}  # Add more punctuation as needed
     progress = tqdm(gen_text_batches)
-    ipa_text_ref = text_to_ipa(ref_text, language=language)
+    ipa_text_ref = text_to_ipa(ref_text, language=ref_language)
+    print(ref_language)
     print(language)
 
     for i, gen_text in enumerate(progress):
@@ -444,7 +446,7 @@ def infer(ref_audio_orig, ref_text, gen_text, exp_name, remove_silence, cross_fa
             ref_audio,
             chunk_length_s=15,
             batch_size=128,
-            generate_kwargs={"task": "transcribe"# ,"language": language  # Use the variable here
+            generate_kwargs={"task": "transcribe"# ,"language": ref_language  # Use the variable here
             },
             return_timestamps=False,
         )["text"].strip()
@@ -552,6 +554,11 @@ def update_language(new_language):
     language = new_language
     return f"Language set to: {language}"
 
+def update_language1(new_ref_language):
+    global ref_language
+    ref_language = new_ref_language
+    return f"Language set to: {ref_language}"
+
 def update_speed(new_speed):
     global speed
     speed = new_speed
@@ -572,8 +579,47 @@ with gr.Blocks() as app_tts:
     model_choice = gr.Radio(
         choices=["Multi"], label="Choose TTS Model", value="Multi"
     )
+    gr.Markdown("#Select Reference Language")
+    language_choice1 = gr.Dropdown(
+        choices=["pl", "de", "en-us", "en-gb", "uk", "ru", "cs",  # Czech
+    "sk",  # Slovak
+    "bg",  # Bulgarian
+    "sr",  # Serbian
+    "hr",  # Croatian
+    "sl",  # Slovenian
+    "be",  # Belarusian
+    "lt",  # Lithuanian
+    "lv",  # Latvian
+    "et",  # Estonian
+    "fi",  # Finnish
+    "hu",  # Hungarian
+    "sv",  # Swedish
+    "no",  # Norwegian
+    "da",  # Danish
+    "is",  # Icelandic
+    "nl"  # Dutch
+                ], label="Choose Language", value="en-us"
+    )
+    gr.Markdown("#Select Synthesized Language")
     language_choice = gr.Dropdown(
-        choices=["pl", "de", "en", "en-us", "en-gb", "uk", "ru"], label="Choose Language", value="en"
+        choices=["pl", "de", "en-us", "en-gb", "uk", "ru", "cs",  # Czech
+    "sk",  # Slovak
+    "bg",  # Bulgarian
+    "sr",  # Serbian
+    "hr",  # Croatian
+    "sl",  # Slovenian
+    "be",  # Belarusian
+    "lt",  # Lithuanian
+    "lv",  # Latvian
+    "et",  # Estonian
+    "fi",  # Finnish
+    "hu",  # Hungarian
+    "sv",  # Swedish
+    "no",  # Norwegian
+    "da",  # Danish
+    "is",  # Icelandic
+    "nl"  # Dutch
+                ], label="Choose Language", value="en-us"
     )
     generate_btn = gr.Button("Synthesize", variant="primary")
     with gr.Accordion("Advanced Settings", open=False):
@@ -605,6 +651,7 @@ with gr.Blocks() as app_tts:
         )
     speed_slider.change(update_speed, inputs=speed_slider)
     language_choice.change(update_language, inputs=language_choice)
+    language_choice1.change(update_language1, inputs=language_choice1)
 
 
     audio_output = gr.Audio(label="Synthesized Audio")